Computational methods for the study of enzymic reaction mechanisms. 1. Application to the hydride transfer step in the catalysis of dihydrofolate reductase

We report results and assessment from the use of several "state-of-the-art" computational methods to investigate the effect of the Asp-27 ionization state on the hydride-ion transfer step in the enzymic reduction of folate and dihydrofolate (DHF) by DHFR from Escherichia coli (E. coli). The active forms of the DHFR complex are assumed protonated on the pterin/dihydropterin ring of the folate/dihydrofolate molecule prior to transfer of the hydride ion from the nicotinamide adenine dinucleotide phosphate (NADPH) cofactor. The calculations have been carried out for both protonated (neutral) and unprotonated (negatively charged) states of the conserved active-site Asp27 (E. coli) residue in the DHFR complexes. First, geometry optimizations at the semiempirical (PM3), density functional (DFT) and ab initio levels were performed on reaction-analogue clusters to obtain reactant, transition state (TS), and product complexes. The DFT and ab initio reaction energies obtained for the unprotonated Asp complexes were endothermic: exothermic reaction energies were obtained only for the protonated Asp complexes. In contrast, the PM3 method wrongly predicted endothermic reactions for both unprotonated and protonated Asp. Thus, for these simple analogue systems, the results suggest that reduction takes place only if the active-site Asp27 is protonated. Second, the effects of the remaining substrate, cofactor, protein and solvent interactions were studied using hybrid semiempirical (PM3) quantum mechanical and molecular mechanics (QM/MM) methods combined with molecular dynamics (MID) simulations. Within the MD simulation scheme, TS complexes were characterized using an efficient implementation of the coordinate-driving technique. Errors arising from the use of PM3 in the QM/MM calculations were estimated and corrected using the results of additional cluster calculations at the DFT and MP2 levels. Contrary to the initial cluster calculations, these corrected QM/MM calculations did not indicate unambiguously that reduction takes place only if the active-site Asp27 is protonated. For both ionized and neutral states of Asp27, analysis of the electrostatic interaction energies between the QM and MM regions in the MD simulations showed that the enzymic environment (MM region) plays a far greater role in stabilizing the final products than in stabilizing the TS complex. Activation free energies for the hydride-ion transfer ranged from 10 to 30 kcal/mol depending on the choice of QM region and ionization state of Asp27. For the larger QM region, the range of activation free energies (10-12 kcal/mol) is similar to an experimental value (13 kcal/mol), but in general quantitative agreement with experimental free energies was poor. Comparative estimates of reaction free energies suggest the catalytic advantage of a protonated Asp27 residue is small. Methodologically, the present work has identified issues relating to the composition and semiempirical (PM3) treatment of the QM region which need to be addressed in future studies.